Automatic Sales and Order Forecasting at WINETIME: Insights from Logistics Director Ruslan Sheptytskyi

WINETIME is a Ukrainian chain of wine and gastro markets operating in 14 cities across Ukraine. A year ago, the company reconfigured its auto-ordering system and integrated the MySales forecasting system. As a result, forecasting accuracy improved by 10–15%, while non-productive inventory decreased by 3–5%, depending on the category. Store replenishment is carried out automatically, taking into account seasonality, promotional campaigns, and other factors affecting sales. Thanks to this, the number of errors caused by the human factor has significantly decreased. Ruslan Sheptytskyi, Logistics Director at WINETIME, spoke about this.

— Please tell our readers about the WINETIME business. How many stores are in the chain and what format do you operate in?

— WINETIME is a chain of gastro and wine markets in 14 cities of Ukraine. It is a community of like-minded people united by one goal — to fill life with new delicious experiences.

Currently, the chain includes 31 gastro markets. WINETIME is represented in various formats: from compact stores of 100 sq. m to markets with an area of 1,200 sq. m. Many of the markets feature food halls where guests can relax and enjoy their favorite French, Mediterranean, and Japanese dishes. Guests can also ask for any products from the shelves to be professionally prepared.

The assortment is based on direct imports from dozens of countries of the Old and New Worlds, a huge selection of vegetables and exotic fruits, delicacy meats and fish, products from local farmers, and freshly baked bread from a wood-fired oven. In addition, there is a dedicated area for healthy food, lactose-free, and vegan products.

We actively cooperate with 600 alcohol producers and 200 food manufacturers. All production facilities undergo mandatory quality control by our internal audit team so that both we and our guests can be confident in the undeniable quality and freshness of the products placed on the shelves. In addition to the retail chain, the business also includes the TROSTYNKA farming enterprise and the ZHORNOVA craft bakery.

— In which price segment do you operate?

— The WINETIME gastro market chain operates in the middle, “middle-plus,” and premium segments. In our stores, customers can find products both for everyday consumption and for festive occasions.

— What factors influence sales in your business, and how are they taken into account in forecasting and auto-ordering?

— As in all retail, the main factors influencing sales are product seasonality, activities in related categories within other retail chains, and promotional activities carried out within WINETIME itself.

All these factors are taken into account by the MySales forecasting system, which we recently implemented into our supply chain model and integrated with our proprietary auto-ordering system.

— How often do you conduct promotional campaigns?

— As a rule, we have both long-term planning for significant calendar dates such as International Women’s Day, Black Friday, New Year, etc., as well as short-term promotional offers for different product categories every month.

— What share of sales comes from promotional offers? By which indicators do you evaluate their effectiveness?

— It varies depending on the period. In general, this share fluctuates between 10% and 15%, depending on the promotional period and the marketing activities agreed upon with manufacturers.

We evaluate the effectiveness of promotions based on sales growth indicators, the number of receipts, and the economic effect of invested resources compared to the resulting revenue.

— What principles were used to build the order forecasting system?

— Before we started working with the MySales forecasting system, we used our own modifications in the auto-ordering module along with the expert assessments of our head sommeliers in stores. To some extent and for some time, this satisfied our needs, but with the growth in the number of stores, as well as the expansion of the assortment and the number of suppliers we worked with, this became a complex task that did not always produce the expected results. Therefore, we began searching for a more advanced solution for sales and order forecasting.

— What task did you approach MySales with?

— We approached the MySales team specifically with the task of forecasting replenishment for our retail locations, which ultimately led to the reconfiguration of the existing auto-ordering system. Our main need was to improve quality, speed, and the ability to work with large volumes of replenishment data for both regular and promotional sales, including fresh products.

Also read: How the “Chudo Market” chain increased on-shelf availability to 97% and improved inventory turnover by 20%

— What changed in the forecasting and auto-ordering processes after implementing MySales?

— We decided not to develop a forecasting block within our own auto-ordering module and instead chose to connect an external system that would provide more accurate data. To achieve this, we searched for potential contractors capable of offering the required solution. We ultimately chose the MySales system.

During the testing process, which lasted about a year, we changed the structure of our inventory management system. It now consists of the MySales forecasting system, our proprietary auto-ordering module, and the transportation system. All of them are integrated into a single structure aimed at ensuring accurate and high-quality replenishment of our stores.

— How closely do the forecasting indicators match actual sales?

— We launched the MySales forecasting system relatively recently, but the forecasting quality achieved during the first months of industrial use already reached 75–80%. Before that, this figure fluctuated at around 60–65%. As a result, we increased product availability for our guests. Our goal for the next six months is to achieve forecasting accuracy of over 90% for the chain’s main categories.

— For what period is the forecast built?

— We carry out forecasting for different time horizons depending on the purpose. For example, some forecasts are built six months ahead, while others cover the upcoming week.

Long-term forecasting is used to form our orders for imported products. Since we deal with goods from other continents and seasonal wine production, fulfillment of such orders can take up to six months. Shorter-term forecasting is necessary for replenishing stores from the distribution center.

— How often and according to what principles are inventories replenished in stores and warehouses?

— In the warehouse and throughout the network in general, replenishment is based on the EOQ (Economic Order Quantity) principle, unless there are specific contractual conditions with manufacturers. We calculate the batch volume relative to logistics costs and storage financing costs against the profit generated from sales.

In stores, we focus more on product representation on shelves according to assortment matrices and planograms, as well as on sales between delivery cycles from the distribution center.

— How long did the MySales integration process take? Which IT systems required integration?

— The integration itself did not take long. Within several months, we configured the required data exchange, servers, and data structures. More time was needed for adaptation and adjustments related to different assortment specifics, supply cycles, data for various types of promotions, and the actual testing process at several retail locations within the chain.

Overall, it took about nine months to achieve a full launch with a properly configured system and complete integration with our proprietary auto-ordering system.

— What integration results of MySales could you share?

— We now manage forecasts for virtually all categories. Currently, the process of configuring and adapting the system for super-fresh products with shelf lives of up to seven days is ongoing.

Our forecasting accuracy level has improved by 10–15% and continues to grow. Inventory turnover increased by 3–4%, which is significant considering long import lead times. Non-productive inventory decreased by 3–5%, depending on the category.